Importance of M-Protein C Terminus as Substrate Antigen for Serodetection of Equine Arteritis Virus Infection

Abstract

Equine arteritis virus (EAV), an enveloped positive-stranded RNA virus, is the prototype of the arterivirus group. In a previous paper (A. Kheyar, S. Martin, G. St.-Laurent, P. J. Timoney, W. H. McCollum, and D. Archambault, Clin. Diagn. Lab. Immunol. 4:648-652, 1997), we have shown that the unglycosylated membrane (M) protein, which is composed of 162 amino acids (aa), is a major target of equine antibody to EAV. In order to determine the antigenic regions of the M protein, the cDNA encoding the M protein of EAV was inserted into the procaryotic expression vector pGEX-4T-1 to produce recombinant glutathione S-transferase-M fusion protein. Various deletion mutant clones, which covered the entire sequence of the M protein, were then generated by inverse PCR and expressed in Escherichia coli to examine, by a Western blot assay, the antigenic reactivity of the clone-derived truncated M proteins with sera from horses either experimentally or naturally infected with EAV. Deletion of the hydrophobic N-terminal 87 aa did not abolish immune reactivity of the protein with serum antibodies to EAV, thereby demonstrating the antigenicity of the C-terminal region (aa 88 to 162) of the M protein. Further truncations of the M-protein C-terminal domain defined particular linear epitope-containing amino acid sequence regions. However, only the M-protein C-terminal region was readily recognized by all EAV-specific horse antisera tested in this study. Based on these findings, only the M-protein C-terminal polypeptide composed of aa 88 to 162 is necessary to identify horse serum antibodies specific to the EAV M protein. Thus, this polypeptide might be useful for serodetection of EAV-infected animals.

Equine arteritis virus (EAV) is the causative agent of equine viral arteritis, a contagious viral infection of equids (16, 34). The clinical outcome following EAV exposure of horses varies from subclinical infection to systemic EAV disease, which may result in abortion by pregnant mares. A high percentage (30 to 60%) of the stallions infected with EAV become persistently infected long-term carriers and, consequently, play an important role in perpetuation and venereal dissemination of the virus (34).

EAV is the prototype member of the family Arteriviridae in the order Nidovirales together with lactate dehydrogenase-elevating virus, porcine reproductive and respiratory syndrome virus (PRRSV), and simian hemorrhagic fever virus (5). The EAV genome is a positive, single-stranded, polyadenylated RNA molecule of 12.7 kb in length (12). It contains, in the direction 5′-3′, two large open reading frames (ORFs), 1a and 1b, which represent approximately three-quarters of the genome, and seven smaller ORFs designated 2a, 2b, and 3 to 7 (12, 32). During cell infection, ORFs 2a, 2b, and 3 to 7 are expressed as a nested set of leader-containing subgenomic viral mRNAs (12, 14). ORFs 1a and 1b encode the viral replicase, whereas the known EAV structural proteins E (8 kDa), G_S (25 kDa), G_L (30 to 42 kDa), M (16 kDa), and N (14 kDa) are encoded by ORFs 2a, 2b, 5, 6, and 7, respectively (15, 32). Finally, the products encoded by ORFs 3 and 4 are glycosylated membrane-associated proteins, the functional role of which is still under debate (15, 21).

The diagnosis of EAV infection is currently based on virus isolation in cell cultures and/or EAV-specific antibody detection in sera of infected animals (34). Although enzyme-linked immunosorbent assays (ELISAs) in which whole virions; recombinant G_L, M, and/or N proteins; or ovalbumin-conjugated G_L-specific synthetic peptide was used as the test antigen have been reported previously (8, 9, 10, 20, 28), the serum neutralization (SN) test, which detects antibodies to the G_L glycoprotein, is the assay currently recognized as the international standard test for determination of the serological status of horses infected with EAV (30). However, the SN test, although reliable, is relatively expensive and laborious, and it takes days to obtain results. In addition, antigenic differences are more likely to be found in the EAV G_L protein, which expresses the neutralizing determinants (2, 3, 7, 13, 19, 33). Thus, to determine the presence of EAV antibodies in the serum of infected horses, it is relevant to search for antibodies which are specific to conserved amino acid regions of EAV proteins. Because high degrees of amino acid sequence homology have been reported previously for M and N proteins of geographically distinct EAV isolates (6), these viral proteins represent suitable candidates to be used as test antigens in a serological assay to detect EAV-infected horses.

Analyses of the humoral immune responses of horses elicited during natural and experimental EAV infections have shown that the M protein is the EAV structural protein most consistently recognized by sera from these animals (20, 24). Although the M protein is a suitable antigen to be used for serological diagnosis of EAV infection, the M-protein antibody-binding regions have yet to be determined. The purpose of this study was to identify the antigenic regions of the EAV M protein by using various deletion mutants that were generated, by the inverse PCR (iPCR) approach, from the wild-type (wt) EAV M-protein-encoding ORF 6. The resulting truncated M proteins produced in a procaryotic expression system were analyzed in an immunoblotting procedure by using sera from horses either naturally or experimentally infected with EAV. The results demonstrated the existence of a strongly antigenic region located in the C-terminal half of the M protein that was readily recognized by all EAV-specific horse antisera tested in this study.

MATERIALS AND METHODS

Construction of pGex-ORF6 recombinant plasmid and deletion mutant clones.

The cloning of EAV ORF 6 coding sequence (derived from cell-passaged EAV Bucyrus strain [16]) into the pCR II TA cloning vector has been previously described (23). The ORF 6 cDNA fragment was excised from the pCR II TA vector with the appropriate restriction enzymes, purified by using a low-melting-temperature agarose gel, and ligated into the procaryotic expression vector pGEX-4T-1 (Amersham Pharmacia Biotech, Baie d'Urfé, Quebec, Canada). This procedure allowed the ORF 6 encoding the EAV M protein to be in frame with the glutathione S-transferase (GST) gene, generating the recombinant plasmid pGex-ORF6 which then could express the GST-M fusion protein. The recombinant plasmid was sequenced by the chain termination method (29) to confirm that the junction sequence was in the appropriate reading frame.

Deletion mutant plasmid clones derived from pGex-ORF6 were generated by iPCR (22) using two 5′-phosphorylated primers in inverted tail-to-tail directions to amplify the entire pGex-ORF6 plasmid, except for each ORF 6 coding region to be deleted. The ORF 6 primer pairs were selected according to the published sequence of the EAV Bucyrus strain genome (12) and/or the ORF 6 sequence obtained above. The nucleotide sequences of the primers and the DNA fragments used as templates for iPCR amplification are shown in Table 1. All iPCRs were performed by using Pfu DNA polymerase (Stratagene, La Jolla, Calif.), which exhibits 3′→5′ exonuclease proofreading activity and generates blunt-ended amplification products. The resulting iPCR products were recircularized by self-ligation with the T4 DNA ligase and then used to transform Escherichia coli DH5α. The deletion mutant clones (pGex-ORF6m) were sequenced across the deletion junctions as described above to localize the sites of deletion and to confirm that the deletion mutant clones were in the appropriate reading frame for expression.

TABLE 1.

Nucleotide sequences of primers used in iPCR to generate the EAV M mutant proteins

M-protein fragmenta	Sequenceb	Positions (nt)c	Template pGex-ORF6d
A	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	1-489
	(−) 5′ pCATACCTACAATCATCCTCGT 3′	261 to 241
B	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	1-262
	(−) 5′ pATCTAGATACTCACCTAAAAT 3′	54 to 34
C	(+) 5′ pATGCCTCGTCTTCGGTCCATT 3′	262 to 282	1-489
	(−) 5′ pGGATCCACGCGGAACCAGATC 3′	−1 to −21
D	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	262-489
	(−) 5′ pGGTGTACCCGTTGCCGCGAAC 3′	390 to 370
E	(+) 5′ pACCGCAGTTGGTAACAAGCTT 3′	388 to 408	262-489
	(−) 5′ pGGATCCACGCGGAACCAGATC 3′	−1 to −21
F	(+) 5′ pGGCGTCAAGACGATCACGTCC 3′	415 to 435	262-489
	(−) 5′ pGGATCCACGCGGAACCAGATC 3′	−1 to −21
G	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	262-390
	(−) 5′ pGTCCACAAAATCAGCTACCAC 3′	321 to 301
H	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	313-489
	(−) 5′ pCTGAGTAGTTGAGCGGGGGAT 3′	363 to 343
I	(+) 5′ pACACCTAGTGGACCTGTTCCC 3′	322 to 342	262-390
	(−) 5′ pGGATCCACGCGGAACCAGATC 3′	−1 to −21
J	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	361-489
	(−) 5′ pGCCATCGACAAGCTTGTTACC 3′	417 to 397
K	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	387-489
	(−) 5′ pGCCTGCGGACGTGATCGTCTT 3′	441 to 421
L	(+) 5′ pTGACCTACTGCGCCTGCAGGA 3′	487 to +18	415-489
	(−) 5′ pCGCCATCCGTTTCGAACAGAG 3′	465 to 445
M	(+) 5′ pCGCCTCTGTTCGAAACGGATG 3′	442 to 462	387-489
	(−) 5′ pGGATCCACGCGGAACCAGATC 3′	−1 to −21

^aThe amino acid positions of individual M-protein fragments are shown in Fig. 1.

^bAll primers are 5′ phosphorylated (p); (+), sense primers; (−), antisense primers.

^cThe nucleotide (nt) position corresponds to the first nucleotide of the EAV ORF 6 coding sequence cloned into pCR II TA vector. The negative or positive number corresponds to the number of additional nucleotides cloned into the pGex expression vector which are not on ORF 6.

^dPlasmid constructs used as templates for iPCR amplification. The nucleotide positions of the coding regions of ORF 6 deletion mutants are indicated.

Expression and purification of fusion proteins.

The procedures used for the expression and purification of intact and truncated M fusion proteins were similar to those employed in our laboratory (1, 23). Briefly, cultures of E. coli DH5α containing the parental plasmid pGex-ORF6, or each of the generated deletion mutants, were grown in 2× yeast extract-tryptone medium containing ampicillin (200 μg/ml) and induced with isopropyl-β-d-thiogalactoside (IPTG; 0.1 mM) for 4 h at 37°C. The resulting bacterial cells were pelleted, resuspended in loading buffer, and boiled for 5 min before fractionation by sodium dodecyl sulfate (SDS)-12% polyacrylamide gel electrophoresis. When partially insoluble, fusion proteins were solubilized from inclusion bodies by using N-lauroylsarcosine (Sarkosyl), as described by Frangioni and Neel (17). Fusion proteins were purified either by using the glutathione-Sepharose 4B affinity matrix (Amersham Pharmacia Biotech) or by electroelution of the proteins (Microeluter; Bio-Rad Laboratories, Palo Alto, Calif.) from an SDS-polyacrylamide gel (23). The amount of purified proteins was then judged from SDS-polyacrylamide gels stained with Coomassie brilliant blue.

Sera.

To investigate the reactivity of equine anti-EAV antibodies with the GST-M fusion proteins, 14 EAV-positive horse antisera (kindly provided by William H. McCollum and Peter Timoney, Gluck Equine Research Center, University of Kentucky, Lexington) were tested. Six EAV-positive antisera were from naturally EAV-infected horses while the other eight were from horses that were convalescent from experimental EAV infection with the reference Bucyrus strain or the field isolate KY84, IL-93, or IL-94 (23). As an additional control, another serum sample positive for EAV antibodies (generously provided by Claude Dubuc, Virology Section, Animal Diseases Research Institute, Canadian Food Inspection Agency, Nepean, Ontario, Canada) was obtained from a horse at day 96 after experimental infection with the EAV Bucyrus reference strain. All horse serum samples used in this study were initially tested by the SN assay for the presence of EAV neutralization antibodies (30). The SN titers of the various EAV-positive sera ranged from 4 to ≥512. As negative controls, six anti-EAV-negative horse sera (provided by William H. McCollum and Peter Timoney) and field antisera from horses naturally infected with equine herpesvirus type 1 and equine influenza virus type 1 (provided by Susan Carman, Veterinary Laboratory Services, Ontario Ministry of Agriculture Food and Rural Affairs, Guelph, Ontario, Canada) were used. A porcine serum anti-PRRSV antibody (a gift from Ronald Magar, Health of Animals and Food Laboratory, Canadian Food Inspection Agency, St.-Hyacinthe, Quebec, Canada) was used as an additional negative control.

Immunoreactivity of the fusion proteins by Western immunoblotting.

The immunological reactivity of the horse antisera to the fusion M proteins was determined by Western immunoblotting. Purified proteins were separated by SDS-12% polyacrylamide gel electrophoresis as described above and then electrotransferred onto nitrocellulose membranes. Immunoblotting was performed as described previously (23) by using, as the blocking reagent solution, 5% nonfat dried milk solids and 0.05% Tween 20 in phosphate-buffered saline solution, pH 7.5. The horse serum samples were used at a final dilution of 1:50 in the blocking reagent solution, while a peroxidase-conjugated rabbit anti-horse immunoglobulin G (whole molecule) was used as a secondary antibody. Where appropriate, preincubation of the diluted horse antisera for 90 min at 37°C with an excess of purified GST (0.1 μg/ml) was carried out prior to the immunoblotting to reduce background staining due to the reactivity of certain horse sera to the GST portion of the fusion protein (1).

RESULTS

Generation and analysis of various M-protein fragments expressed in E. coli.

Full-length and truncated M proteins of the EAV were produced in the pGEX-4T-1 expression vector as GST fusion proteins. The EAV ORF 6 coding sequence (GenBank accession no. AF320572) was successfully inserted into the pGEX-4T-1 vector where the tac promoter could be adequately controlled by IPTG. Thirteen deletion mutants of the recombinant plasmid pGex-ORF6, which covered the entire M-protein-encoding sequence, were then generated by iPCR in order to express several ORF 6-encoded overlapping fragments. In addition, certain plasmid mutant constructs were also designed by taking into consideration the predicted hydropathy profile of the EAV M protein (15). The length of the M-protein fragments (A to M) (excluding the GST partner) varied from 15 to 87 amino acids (aa) (Fig. 1).

FIG. 1.

Schematic diagram of the structure of the M-protein plasmid constructs. The amino acid sequence of the EAV M protein with the triple membrane-spanning regions (TM) (15) (shaded boxes) is shown at the top. Below is a schematic representation of the full-length protein (wt) and various M-protein deletion constructs. The names (wt and A to M) and amino acid positions of individual M fragments are also shown.

The GST-M fusion proteins expressed in E. coli were obtained from the wt and each mutant clone, purified, and analyzed by Coomassie brilliant blue staining of SDS-polyacrylamide gels (Fig. 2). The levels of expression obtained for the wt M protein and the larger M mutant protein fragment A were reproducibly less than those obtained for the other fusion protein fragments (data not shown). This lower degree of protein expression was believed to be related to the predicted low solubility index of the M-protein N terminus (aa 1 to 87), as suggested by the predominant presence of three highly hydrophobic domains in this region of the protein (15). Because the larger M-protein fragments (wt, A, and C) were insoluble, these fusion proteins were electropurified. In addition, fragments D and G were also found to be mostly insoluble, confirming the predicted hydrophobic profile of aa 100 to 105 (data not shown). In contrast, the M-protein fragments B, E, F, and H to M, predicted to show a hydrophilic profile, could be easily solubilized.

FIG. 2.

Expression of GST and GST-M fusion proteins as analyzed by SDS-polyacrylamide gel electrophoresis. Each lane represents purified GST and GST-M proteins (wt and A to M fragments) stained with Coomassie brilliant blue. Molecular mass standards (kilodaltons) are indicated on the left.

Figure 2 shows the protein bands (wt M and fragments A to M) which were obtained following IPTG induction and purification. These recombinant proteins were larger than the GST (29-kDa) fusion partner with size increases over the native GST ranging approximately from 2 to 14 kDa. The sizes of the protein bands observed on SDS-polyacrylamide gels were in agreement with the predicted molecular mass for each expressed protein, except for the full-length M and the M-protein hydrophobic N-terminal fragment A (aa 1 to 87) (believed to contain the transmembrane domains of the M protein), whose molecular weights were less than predicted. The results have also shown an additional purified band that was consistently observed below the M-protein fragment I-associated major band (aa 108 to 130). This lower band could probably refer to a partially degraded fusion protein or could have been produced by an early termination event of translation (25).

Antigenicity of the GST-M fusion proteins as determined by Western immunoblotting.

The antigenicity of EAV recombinant wt M protein and the various fragments of M protein expressed as GST fusion proteins was investigated by Western immunoblotting. To do this, each fusion protein was allowed to react with EAV-specific horse antisera. Figure 3 shows the results obtained with sera of two different horses that were either experimentally (with the Bucyrus reference strain) or naturally infected with EAV. Both horse antisera readily recognized the full-length (wt) M fusion protein. These EAV M-protein-positive horse antisera did not react with the GST fusion partner alone, thereby showing the specificity of the antibody binding to the EAV M protein (data not shown). Deletion of the hydrophobic N-terminal 87 aa did not abolish immune reactivity of the resulting mutant protein (fragment C) with the EAV horse antisera, thereby demonstrating the antigenicity of the C-terminal region of the M protein. In contrast, no immune reactivity was obtained when the M mutant protein region spanning aa 1 to 87 (fragment A) was allowed to react with these EAV-specific horse antisera. In addition, no immune reactivity was obtained with fragment B (aa 1 to 18), a hydrophilic portion located in the M-protein extreme N-terminal region and believed to be exposed at the surface of EAV particles. However, the EAV-positive horse serum samples readily recognized the C-terminal truncation mutant fragments C to F but not fragment G, thereby demonstrating the presence of linear epitopes within the C-terminal region (aa 108 to 162) of the M protein. Finally, all EAV M-specific fusion proteins (fragments H to M) encompassing the C-terminal amino acid sequence (aa 105 to 162) were recognized by the naturally infected horse antiserum (SN titer, ≥512), whereas three of these fragments (H, K, and M) displayed no reactivity with the serum sample (SN titer, 256) from the experimentally EAV-infected horse.

FIG. 3.

Immune reactivity of GST-M fusion proteins with horse anti-EAV sera by immunoblotting assay. Each panel represents the reactivity of the EAV wt and mutant (fragments A to M) M proteins to EAV horse antisera. Sera from a horse experimentally infected with the EAV Bucyrus reference strain (SN titer, 256) and from a horse naturally infected with EAV (SN titer, ≥512) were used at a 1:50 dilution. Molecular mass standards (kilodaltons) are indicated on the left.

To investigate further the antigenicity of the M protein, different fragments of the carboxy-terminal region of the EAV M protein, which appeared to contain the M-protein linear epitopes, were probed with a panel of sera from naturally or experimentally EAV-infected horses. These serum samples were selected from horses seropositive (by the SN test) for EAV with neutralizing antibody titers varying from 4 to ≥512. All horse EAV antisera that were tested reacted similarly in magnitude with the C-terminal region C fragment (aa 88 to 162) of the M protein, regardless of the EAV-specific neutralizing antibody titers (Table 2). Although there was variation in the degree of signal reactivity for the other fragments (data not shown), the results showed that the E fragment (aa 130 to 162) immunoreacted with 93% (14 out of 15) of the horse serum samples. The J fragment (aa 121 to 139) was found to be the most immunoreactive of the shorter M-protein C-terminal fragments described in this study, as shown by its ability to react with 80% of the EAV horse antisera. In contrast, fragments G (aa 88 to 107) and M (aa 148 to 162) were the least immunoreactive regions of the M-protein fragments, reacting with less than 15% of the sera tested. The immunoblotting results also showed that all but one EAV antibody-containing horse antiserum reacted with one or more of the fragments I, J, K, and/or L (spanning aa 108 to 155) of the M protein.

TABLE 2.

Horse immune reactivity to various protein fragments targeting the C-terminal region of the EAV M protein by immunoblotting assay

Serum no.	SN titerc	EAV infectiond	Immune reactivitya with C-terminal region of the M proteinb
			C	D	E	F	G	H	I	J	K	L	M
4	≥512	Natural	+	−	+	−	−	−	−	+	+	−	−
7	≥512	Natural	+	+	+	+	−	+	+	+	+	+	+
8	≥512	Natural	+	−	+	−	−	−	−	−	+	−	−
9	≥512	Natural	+	+	+	−	−	−	+	+	+	−	−
16	32	Natural	+	−	+	−	−	−	−	+	+	−	−
18	4	Natural	+	+	+	−	−	−	+	+	−	−	−
12	16	KY84	+	−	+	+	−	−	−	+	−	+	−
13	16	KY84	+	−	+	+	−	−	−	+	−	+	−
15	256	Bucyrus	+	+	+	+	−	−	+	+	−	+	−
30	64	KY84 (16 wk)	+	−	+	+	−	−	−	+	+	+	−
2	64	IL-93 (5 wk)	+	+	+	+	−	−	+	−	−	+	−
5	64	IL-93 (5 wk)	+	+	+	+	−	−	+	+	+	+	+
6	64	IL-93 (5 wk)	+	+	+	+	−	+	+	+	+	+	−
59	128	IL-94 (6 wk)	NA	+	−	−	+	+	−	−	−	−	−
X	≥4	96 dpi	+	+	+	+	−	+	−	+	−	−	−

^aImmune reactivity of the fusion M proteins was determined by immunoblotting as described in Materials and Methods. −, no immune reactivity; +, positive immune reactivity; NA, not available.

^bThe amino acid positions of EAV M-protein fragments fused to the GST protein are shown in Fig. 1.

^cSN assay for the presence of EAV neutralization antibodies.

^dSera from naturally EAV-infected horses and from horses that were convalescent from experimental EAV infection with the reference Bucyrus strain or the field isolate KY84, IL-93, or IL-94. The postinfection time when serum was collected from certain experimentally EAV-infected horses is indicated within parentheses. Another equine serum (X) positive for EAV antibodies (96 days postinfection [dpi]) was used. All serum samples from EAV-infected horses were used at a 1:50 dilution.

For controls, no immune reactivity was obtained when the various GST-M fusion proteins targeting the C-terminal region of the M protein were allowed to react with horse sera that were shown to be negative for EAV antibodies by the SN test (data not shown). No cross-reactions were detected when the C-terminal fragments of the M protein were allowed to react with horse antisera specific to equine herpesvirus type 1 and equine influenza virus type 1 and with a porcine antiserum specific to PRRSV (data not shown).

DISCUSSION

The remarkably conserved hydrophobicity profile among nidovirus M proteins suggests that this protein has a similar structure in all members of the order Nidovirales (5). Thus, the nidovirus M protein putatively contains a short amino-terminal domain at the external surface of the virion, three hydrophobic transmembrane segments close to the N-terminal region, and a long carboxy-terminal domain inside the virion, typical of a conventional N-exo, C-endo topology (5). Because of this particular structure, the EAV M protein was believed to be poorly immunogenic (31). However, we demonstrated in an earlier report that the M protein was indeed a potent immunogen in rabbits and was readily recognized by horse sera containing EAV antibodies (23). In addition, studies subsequent to our work have repeatedly shown that the EAV M protein, one of the major structural membrane proteins of the virus, is the protein most consistently recognized by sera from convalescent EAV-infected horses (20, 24). These results and the need for a suitable serological test for the serodetection of EAV-infected horses prompted us to undertake an analysis of M-protein antigenicity in order to define the immunoreactive regions of the protein.

Based on the immune reactivity patterns of horse antisera to EAV with a set of recombinant M deletion mutant proteins, antigenic linear epitope-containing regions were found to be located within the large C-terminal endodomain (aa 88 to 162) of the M protein. In contrast, no linear epitope was found to be located in the putative transmembrane regions (which were contained in fragment A) of the M protein. Furthermore, our results showed that the short hydrophilic N-terminal domain (aa 1 to 18), which is believed to be located at the external surface of the virion and predicted to be immunogenic (31), also lacked immune reactivity to equine EAV-positive sera in the immunoblotting procedure. The lack of immune reactivity of that particular protein fragment with the EAV-positive antisera may be explained by the absence of linear epitopes within this N-terminal domain or by the fact that the antibody response to this putative ectodomain of the M protein might be directed against conformationally dependent epitopes. Further experimentation would be necessary to clarify this point.

All sera tested in this study from experimentally and naturally EAV-infected horses reacted with the carboxy-terminal C fragment (aa 88 to 162) of the M protein. Identical results were found with antisera obtained from rabbits immunized either with maltose-binding protein-M fusion protein (23) or with whole EAV virions (data not shown). The results also have shown that all horse EAV antisera reacted similarly in magnitude with the aa 88 to 162 C fragment of the M protein, independently of the degree of serum EAV-specific neutralizing antibody titers (Table 2). This lack of correlation is not surprising, because the SN test detects antibodies to the G_L glycoprotein, which expresses, as mentioned above, the EAV-neutralizing determinants (2, 3, 7, 13, 19, 33).

Although an immunodominant region located between aa 108 and 155 of the M protein could be identified (on the basis of the immune reactivity of fragments I to L), an individually dependent immune reactivity was obtained with fragments D to M. These results were not surprising, because the humoral immune response to the structural EAV proteins elicited in naturally or experimentally infected horses varies widely with the infecting EAV strain, the interval after infection, and the individual horse (20, 24, 28), indicating that more than one structural protein as substrate antigen would be necessary for serological screening of EAV. It is also noteworthy that variation in the antibody response to EAV structural proteins was also observed for horses immunized with the modified live EAV vaccine (20, 24). Collectively, these latter results and ours are in agreement with those of another report which showed that different antibody responses to the VP2 structural protein of the African horse sickness virus were elicited in horses naturally infected with the virus (4). Although the antigenic variations reported for the M protein of North American and European PRRSV isolates (11, 26) could serve to explain the divergent antibody responses in horses infected with EAV, this mechanism is not likely to be an important factor since the M-protein amino acid sequence is highly conserved among EAV isolates (6).

The use of procaryote-derived fusion proteins has provided a rapid and reliable tool for the determination of the antigenic structure of several viral proteins (27, 35). Here, the iPCR procedure was used to introduce selected deletions within the M-protein amino acid sequence in a single experimental step, avoiding the use of subcloning procedures to generate mutant proteins. It should also be noted that studies with fusion proteins expressed from such deletion clones can provide information only on the approximate location of epitopes within a protein chain due to the limits in the number and size of fusion proteins that can be obtained and analyzed. Thus, further studies are needed to better delineate the precise location and size of the epitopes within the antigenic carboxy terminus of the EAV M protein, by, for instance, generating synthetic overlapping peptides to be used in a Pepscan analysis (18). In addition, since the fragments of the M protein were expressed as bacterial fusion proteins and their antigenicity was analyzed by immunoblotting, our study of the antigenic structure of this protein was limited to linear amino acid sequences.

In summary, a linear epitope-containing region located between aa 108 and 155 of the M protein was identified with the EAV antibody-containing horse antisera tested in this study. However, the region from aa 88 to 162 of the M protein, which was shown to be the only M-protein fragment recognized by all horse EAV-specific antisera tested, would be the peptide of choice to be used as a substrate antigen for the detection of EAV M-protein-specific antibodies. Nevertheless, since a number of horse sera (either negative or positive for EAV) reacted with the GST fusion partner alone and the procedures using the thrombin factor to cleave off the viral proteins from the GST gave poor and inconsistent results (data not shown), it would be necessary to express the selected M-protein fragment in another expression system with no fusion partner for the serodetection of EAV infection.